Results and Demos


During the first year of the project, partners developed a set of showcases (i.e., hardware /software components that demonstrate state-of-the-art technologies available at some partners).

Showcase Responsible Partner Description
BSS + particle filtering FBK Separation of multiple speakers and localization of the sources, based on a distributed microphone network.
SBSS-based AEC FBK Acoustic echo cancellation in a noisy environment based on distributed loudspeakers.
Text understanding FBK Text understanding based on a machine learning approach, including:
- text preprocessing, target identification, frame identification, argument identification
- production of an XML file embedding the text input augmented with frame-semantic information
- related model training
Dialogue mock-up NewAmuser Dialogue management based on a close-talking interaction and on hand-crafted grammar-based understanding.
MEMS-array platform ST Italy Stand-alone system for recording, data transfer to a PC, and processing based on algorithms implemented during the first year.
Distant-talking ASR FBK Speech recognition with microphone arrays under controlled conditions.
ASR in Portuguese INESC-ID Baseline system with close-talking input.
ASR in German TU Graz Baseline system with close-talking input.
Front-end processing  ATHENA-RC-IAMU Techniques performing speech enhancement.